@Edent's new blogpost, "Mastodon Now Sends Referer Headers! Hurrah!", captures the nuances of this issue well, IMHO.
A choice that some server admins may wish to make (hopefully following transparency to their users), off by default, and still a relatively high degree of fragmentation.
Will I do it? Probably not :)
https://shkspr.mobi/blog/2024/12/mastodon-now-sends-referer-headers-hurrah/
@neil @Edent Fuck no. I'm really not affected anyway since I nuke them at the U-A layer, but I hope instances that respect their users won't turn this on. Cc @hachyderm.
@dalias @neil
There are currently over 11,000 accounts on @hachyderm
If they did start providing referer information, what privacy do you think you would lose?
@Edent @neil @hachyderm As I said, I block referer at user agent layer, so none for me. But it's bad policy for the public & web at large, doing something that violates user privacy for the sake of giving publishers what they want, and doing that for the sake of "growth" (promoting Mastodon to them).
@dalias There is nuance here though? _some_ (obviously not you I suppose?) fedi users would like there to be better integrations with publishers (for example, I would prefer that the BBC have their own bots rather than RSS re-publishers), but ️🌈️we live in a society🌈 where you do need to justify doing work, stats help that, and I don't really see a issue if I click a link on mastodon dot social, the BBC knowing that I came from anywhere on mastodon dot social, as @Edent said, there are nuances where you would not want something like that, but generic servers I don't really see the harm, and it does good for a ecosystem (aka, people typically like nice things, this is one of the ways you get nice things)
I just dont understand the threat model of letting the BBC know I came via mastodon.social
@benjojo @Edent "I don't understand why someone wouldn't be okay with having their privacy violated this way" does not make it okay. Research on human subjects requires consent. This includes any kind of market research.
@dalias I feel that a bit of a stretch / bad faith reading of things. Web 'refer' headers have existed for a long time and while they have been curbed in scope (some contexts don't send it at all, some don't send the URL path), it feels a bit extreme to compare this to experimentation on human subjects when if anything the current default was out of the norm